Selective epitaxial formation of semiconductive films

Abstract

Epitaxial layers are selectively formed in semiconductor windows by a cyclical process of repeated blanket deposition and selective etching. The blanket deposition phases leave non-epitaxial material over insulating regions, such as field oxide, and the selective etch phases preferentially remove non-epitaxial material while deposited epitaxial material builds up cycle-by-cycle. Quality of the epitaxial material improves relative to selective processes where no deposition occurs on insulators. Use of a germanium catalyst during the etch phases of the process aid etch rates and facilitate economical maintenance of isothermal and/or isobaric conditions throughout the cycles. Throughput and quality are improved by use of trisilane, formation of amorphous material over the insulating regions and minimizing the thickness ratio of amorphous:epitaxial material in each deposition phase.

Description

FIELD OF THE INVENTION

This invention relates generally to the deposition of silicon-containing materials in semiconductor processing, and relates more specifically to selective formation of silicon-containing materials on semiconductor windows.

BACKGROUND OF THE INVENTION

In forming integrated circuits, epitaxial layers are often desired in selected locations, such as active area mesas among field isolation regions, or even more particularly over defined source and drain regions. While non-epitaxial (amorphous or polycrystalline) material can be selectively removed from over the field isolation regions after deposition, it is typically considered more efficient to simultaneously provide chemical vapor deposition (CVD) and etching chemicals, and to tune conditions to result in zero net deposition over insulative regions and net epitaxial deposition over exposed semiconductor windows. This process, known as selective epitaxial CVD, takes advantage of slow nucleation of typical semiconductor deposition processes on insulators like silicon oxide or silicon nitride. Such selective epitaxial CVD also takes advantage of the naturally greater susceptibility of amorphous and polycrystalline materials to etchants, as compared to the susceptibility of epitaxial layers.

Examples of the many situations in which selective epitaxial formation of semiconductor layers is desirable include a number of schemes for producing strain. The electrical properties of semiconductor materials such as silicon, germanium and silicon germanium alloys are influenced by the degree to which the materials are strained. For example, silicon exhibits enhanced electron mobility under tensile strain, and silicon germanium exhibits enhanced hole mobility under compressive strain. Methods of enhancing the performance of semiconductor materials are of considerable interest and have potential applications in a variety of semiconductor processing applications. Semiconductor processing is typically used in the fabrication of integrated circuits, which entails particularly stringent quality demands, as well as in a variety of other fields. For example, semiconductor processing techniques are also used in the fabrication of flat panel displays using a wide variety of technologies, as well as in the fabrication of microelectromechanical systems (“MEMS”).

A number of approaches for inducing strain in silicon- and germanium-containing materials have focused on exploiting the differences in the lattice constants between various crystalline materials. For example, the lattice constant for crystalline germanium is 5.65 Å, for crystalline silicon is 5.431 Å, and for diamond carbon is 3.567 Å. Heteroepitaxy involves depositing thin layers of a particular crystalline material onto a different crystalline material in such a way that the deposited layer adopts the lattice constant of the underlying single crystal material. For example, using this approach strained silicon germanium layers can be formed by heteroepitaxial deposition onto single crystal silicon substrates. Because the germanium atoms are slightly larger than the silicon atoms, but the deposited heteroepitaxial silicon germanium is constrained to the smaller lattice constant of the silicon beneath it, the silicon germanium is compressively strained to a degree that varies as a function of the germanium content. Typically, the band gap for the silicon germanium layer decreases monotonically from 1.12 eV for pure silicon to 0.67 eV for pure germanium as the germanium content in the silicon germanium increases. In another approach, tensile strain is provided in a thin single crystalline silicon layer by heteroepitaxially depositing the silicon layer onto a relaxed silicon germanium layer. In this example, the heteroepitaxially deposited silicon is strained because its lattice constant is constrained to the larger lattice constant of the relaxed silicon germanium beneath it. The tensile strained heteroepitaxial silicon typically exhibits increased electron mobility. In both of these approaches, the strain is developed at the substrate level before the device (for example, a transistor) is fabricated.

In these examples, strain is introduced into single crystalline silicon-containing materials by replacing silicon atoms with other atoms in the lattice structure. This technique is typically referred to as substitutional doping. For example, substitution of germanium atoms for some of the silicon atoms in the lattice structure of single crystalline silicon produces a compressive strain in the resulting substitutionally doped single crystalline silicon material because the germanium atoms are larger than the silicon atoms that they replace. It is possible to introduce a tensile strain into single crystalline silicon by substitutional doping with carbon, because carbon atoms are smaller than the silicon atoms that they replace. Additional details are provided in “Substitutional Carbon Incorporation and Electronic Characterization of Si_1-yC_y/Si and Si_1-x-yGe_xC_y/Si Heterojunctions” by Judy L. Hoyt, Chapter 3 in “Silicon-Germanium Carbon Alloy”, Taylor and Francis, pp. 59-89 (New York 2002), the disclosure of which is incorporated herein by reference, and is referred to herein as “the Hoyt article.” However, non-substitutional impurities will not induce strain.

Similarly, electrical dopants should also be substitutionally incorporated into epitaxial layers in order to be electrically active. Either the dopants are incorporated as deposited or they will need to be annealed to achieve the desired level of substitutionality and dopant activation. In situ doping of either impurities for tailored lattice constant or electrical dopants are often preferred over ex situ doping followed by annealing to incorporate the dopant into the lattice structure because the annealing consumes thermal budget. However, in practice in situ substitutional doping is complicated by the tendency for the dopant to incorporate non-substitutionally during deposition, for example, by incorporating interstitially in domains or clusters within the silicon, rather than by substituting for silicon atoms in the lattice structure. Non-substitutional doping complicates, for example, carbon doping of silicon, carbon doping of silicon germanium, and doping of silicon and silicon germanium with electrically active dopants. As illustrated in FIG. 3.10 at page 73 of the Hoyt article, prior deposition methods have been used to make crystalline silicon having an in situ doped substitutional carbon content of up to 2.3 atomic %, which corresponds to a lattice spacing of over 5.4 Å and a tensile stress of less than 1.0 GPa.

SUMMARY OF THE INVENTION

In accordance with another aspect of the invention a method is provided for selectively forming semiconductor material in semiconductor windows. The method includes providing a substrate within a chemical vapor deposition chamber, where the substrate comprises insulating surfaces and single-crystal semiconductor surfaces. Semiconductor material is blanket deposited over the insulating surfaces and the single-crystal semiconductor surfaces of the substrate, such that a thickness ratio of non-epitaxial semiconductor material over the insulating surfaces to epitaxial semiconductor material over the single-crystal semiconductor surfaces is less than about 1.6:1. Non-epitaxial semiconductor material is selectively removed from over the insulating surfaces, wherein blanket depositing and selectively removing are conducted within the chemical vapor deposition chamber.

In accordance with another aspect of the invention a method is provided for selectively forming epitaxial semiconductor material. Semiconductor material is blanket deposited to form epitaxial material over single-crystal semiconductor regions of a substrate and to form non-epitaxial material over insulating regions of the substrate. The non-epitaxial material is selectively removed from over the insulating regions by exposing the blanket deposited semiconductor material to an etch chemistry including a halide source and a germanium source. Blanket depositing and selectively removing are repeated at least once.

In accordance with another aspect of the invention a method is provided for forming silicon-containing material in selected locations on a substrate. The method includes providing a substrate having exposed windows of single-crystal semiconductor among field isolation regions. Silicon-containing material is blanket deposited over the windows of single-crystal material and the field isolation regions by flowing trisilane over the substrate. The silicon-containing material is selectively removed from over the field isolation regions. Blanket depositing and selectively removing are repeated in a plurality of cycles.

In accordance with another aspect of the invention a method is provided for selectively forming epitaxial semiconductor material. The method includes providing a substrate with insulating regions and semiconductor windows formed therein. Amorphous semiconductor material is deposited over the insulating regions and the epitaxial semiconductor material is deposited over the semiconductor windows. The amorphous semiconductor material is selectively etched from over the insulating regions while leaving at least some epitaxial semiconductor material in the semiconductor windows. Blanket depositing and selectively removing are repeated in a plurality of cycles.

BRIEF DESCRIPTION OF THE DRAWINGS

Example embodiments of the methods and systems disclosed herein are illustrated in the accompanying drawings, which are for illustrative purposes only. The drawings comprise the following figures, in which like numerals indicate like parts.

FIG. 1 is a flowchart illustrating a process for selectively forming epitaxial semiconductor layers, using the particular example of depositing a carbon-doped silicon film in recessed source/drain regions of a mixed substrate.

FIG. 2 is a schematic illustration of a partially formed semiconductor structure comprising patterned insulator regions formed in a semiconductor substrate.

FIG. 3 is a schematic illustration of the partially formed semiconductor structure of FIG. 2 after performing a blanket deposition of a carbon-doped silicon film over the mixed substrate surface.

FIG. 4 is a schematic illustration of the partially formed semiconductor structure of FIG. 3 after performing a selective chemical vapor etch process to remove carbon-doped silicon from oxide regions of the mixed substrate.

FIGS. 5A-5D are schematic illustrations of the partially formed semiconductor structure of FIG. 4 after performing further cycles of blanket deposition and selective etch.

FIG. 6 shows a graph of etch rate of amorphous regions of a carbon-doped silicon film as a function of HCl partial pressure in the etch chemistry.

FIG. 7 shows a graph of etch rates and ratios amorphous (“a”) and single crystal (“c”) etch rates as a function of GeH₄ flow in the etch chemistry for various etch chemistries.

FIG. 8 shows a graph of etch rate of amorphous regions of a carbon-doped silicon film as a function of chamber pressure.

FIG. 9 shows a graph of etch rate of amorphous regions of a carbon-doped silicon film as a function of reciprocal temperature.

FIG. 10 shows a graph of thickness of amorphous regions of a carbon-doped silicon film as a function of accumulated etch time.

FIG. 11 shows a graph of etch rate of amorphous regions of a carbon-doped silicon film deposited on a wafer as a function of radial position on the wafer.

FIG. 12 shows a graph of thickness of amorphous regions of a carbon-doped silicon film deposited on a wafer as a function of radial position on the wafer for various etch cycle durations.

FIG. 13 shows a graph of thickness of amorphous regions of a carbon-doped silicon film deposited of a wafer as a function of radial position of the wafer for various GeH₄ etchant etch chemistries and various etch cycle durations.

FIG. 14 is a micrograph illustrating an example partially formed carbon-doped silicon structure created by performing blanket deposition and a partial etch cycle on a patterned substrate.

FIG. 15 shows a graph of element concentration as a function of depth for an exemplary partially formed carbon-doped silicon film formed using certain of the techniques disclosed herein.

FIG. 16 is a micrograph illustrating an exemplary formed carbon-doped silicon structure created by performing multiple deposition and etch cycles on a patterned substrate.

FIG. 17 illustrates an atomic force microscopy analysis of an epitaxial carbon-doped silicon film that has been selectively formed using certain of the cyclical techniques disclosed herein.

FIG. 18 shows x-ray diffraction rocking curves of carbon-doped silicon films deposited using certain of the cyclical techniques disclosed herein.

DETAILED DESCRIPTION OF THE INVENTION

Deposition techniques often attempt to tailor the amount or kind of deposition in different regions of a substrate. For example, U.S. Pat. No. 6,998,305 recognizes that simultaneous etch and deposition reactions are know for selective deposition on silicon without depositing on silicon oxide. To control deposition on a third type of surface, namely an exposed transistor gate, the '305 patent teaches cyclically alternating a selective deposition with an etch phase. However, the inventors have recognized that selective deposition chemistries sometimes have undesirably effects on the deposited layers. While the described embodiments involve the specific example of carbon-doped silicon for NMOS applications, the skilled artisan will appreciate that the methods described herein have application to a variety of semiconductor applications where selective formation of a layer is desired but etchants can interfere with desired properties of the deposited layer.

Deposition methods exist that are useful for making a variety of substitutionally doped single crystalline silicon-containing materials. For example, it is possible to perform in situ substitutional carbon doping of crystalline silicon by performing the deposition at a relatively high rate using trisilane (Si₃H₈) as a silicon source and a carbon-containing gas or vapor as a carbon source. Carbon-doped silicon-containing alloys have a complementary nature to silicon germanium systems. The degree of substitutional doping is 70% or greater, expressed as the weight percentage of substitutional carbon dopant based on the total amount of carbon dopant (substitutional and non-substitutional) in the silicon. Techniques for forming carbon-doped silicon-containing materials have overcome several challenges, including the large lattice mismatch between carbon and silicon, the low solubility of carbon in silicon, and the tendency of carbon-doped silicon to precipitate. Additional details relating to in situ substitutional carbon doping of crystalline silicon are provided in U.S. patent application Ser. No. 11/343,275 (filed 30 Jan. 2006).

The term “silicon-containing material” and similar terms are used herein to refer to a broad variety of silicon-containing materials including without limitation silicon (including crystalline silicon), carbon-doped silicon (Si:C), silicon germanium, and carbon-doped silicon germanium (SiGe:C). As used herein, “carbon-doped silicon”, “Si:C”, “silicon germanium”, “carbon-doped silicon germanium”, “SiGe:C” and similar terms refer to materials that contain the indicated chemical elements in various proportions and, optionally, minor amounts of other elements. For example, “silicon germanium” is a material that comprises silicon, germanium and, optionally, other elements, for example, dopants such as carbon and electrically active dopants. Terms such as “Si:C” and “SiGe:C” are not stoichiometric chemical formulas per se and thus are not limited to materials that contain particular ratios of the indicated elements. Furthermore, terms such as Si:C and SiGe:C are not intended to exclude the presence of other dopants, such that a phosphorous and carbon-doped silicon material is included within the term Si:C and the term Si:C:P. The percentage of a dopant (such as carbon, germanium or electrically active dopant) in a silicon-containing film is expressed herein in atomic percent on a whole film basis, unless otherwise stated.

It is possible to determine the amount of carbon substitutionally doped into a silicon-containing material by measuring the perpendicular lattice spacing of the doped silicon-containing material by x-ray diffraction, then applying Vegard's Law by performing a linear interpolation between single crystal silicon and single crystal carbon (diamond). For example, it is possible to determine the amount of carbon substitutionally doped into silicon by measuring the perpendicular lattice spacing of the doped silicon by x-ray diffraction, and then applying Vegard's law. Additional details on this technique are provided in the Hoyt article. It is possible to determine the total carbon content in the doped silicon by secondary ion mass spectrometry (“SIMS”). It is possible to determine the non-substitutional carbon content by subtracting the substitutional carbon content from the total carbon content. It is possible to determine the amount of other elements substitutionally doped into other silicon-containing materials in a similar manner.

“Substrate,” as that term is used herein, refers either to the workpiece upon which deposition is desired, or the surface exposed to one or more deposition gases. For example, in certain embodiments the substrate is a single crystal silicon wafer, a semiconductor-on-insulator (“SOI”) substrate, or an epitaxial silicon surface, a silicon germanium surface, or a III-V material deposited upon a wafer. Workpieces are not limited to wafers, but also include glass, plastic, or other substrates employed in semiconductor processing. As discussed in U.S. Pat. No. 6,900,115, the entire disclosure of which is hereby incorporated by reference herein, a “mixed substrate” is a substrate that has two or more different types of surfaces. For example, in certain applications a mixed substrate comprises a first surface having a first surface morphology and a second surface having a second surface morphology. In certain embodiments, carbon-doped silicon-containing layers are selectively formed over single crystal semiconductor materials while minimizing, and more preferably avoiding, deposition over adjacent dielectrics or insulators. Examples of dielectric materials include silicon dioxide (including low dielectric constant forms such as carbon-doped and fluorine-doped oxides of silicon), silicon nitride, metal oxide and metal silicate. The terms “epitaxial”, “epitaxially” “heteroepitaxial”, “heteroepitaxially” and similar terms are used herein to refer to the deposition of a crystalline silicon-containing material onto a crystalline substrate in such a way that the deposited layer adopts or follows the lattice constant of the substrate. Epitaxial deposition is generally considered to be heteroepitaxial when the composition of the deposited layer is different from that of the substrate.

Even if surfaces are made from the same elements, the surfaces are considered different if the morphologies (crystallinity) of the surfaces are different. The processes described herein are useful for depositing silicon-containing films on a variety of substrates, but are particularly useful for mixed substrates having mixed surface morphologies. Such a mixed substrate comprises a first surface having a first surface morphology and a second surface having a second surface morphology. In this context, “surface morphology” refers to the crystalline structure of the substrate surface. Amorphous and crystalline are examples of different morphologies. Polycrystalline morphology is a crystalline structure that consists of a disorderly arrangement of orderly crystals and thus has an intermediate degree of order. The atoms in a polycrystalline material are ordered within each of the crystals, but the crystals themselves lack long range order with respect to one another. Single crystal morphology is a crystalline structure that has a high degree of long range order. Epitaxial films are characterized by a crystal structure and orientation that is identical to the substrate upon which they are grown, typically single crystal. The atoms in these materials are arranged in a lattice-like structure that persists over relatively long distances (on an atomic scale). Amorphous morphology is a non-crystalline structure having a low degree of order because the atoms lack a definite periodic arrangement. Other morphologies include microcrystalline and mixtures of amorphous and crystalline material. “Non-epitaxial” thus encompasses amorphous, polycrystalline, microcrystalline and mixtures of the same. As used herein, “single-crystal” or “epitaxial” are used to describe a predominantly large crystal structure having a tolerable number of faults therein, as is commonly employed for transistor fabrication. The crystallinity of a layer generally falls along a continuum from amorphous to polycrystalline to single-crystal; a crystal structure is often considered single-crystal or epitaxial, despite low density faults. Specific examples of mixed substrates include without limitation single crystal/polycrystalline, single crystal/amorphous, epitaxial/polycrystalline, epitaxial/amorphous, single crystal/dielectric, epitaxial/dielectric, conductor/dielectric, and semiconductor/dielectric. The term “mixed substrate” includes substrates having more than two different types of surfaces. Methods described herein for depositing silicon-containing films onto mixed substrates having two types of surfaces are also applicable to mixed substrates having three or more different types of surfaces.

When grown into recessed source/drain areas, tensile strained carbon-doped silicon films (Si:C films) provide a tensile strained silicon channel with enhanced electron mobility, particularly beneficial for NMOS devices. This advantageously eliminates the need to provide a relaxed silicon germanium buffer layer to support the strained silicon layer. In such applications, electrically active dopants are advantageously incorporated by in situ doping using dopant sources or dopant precursors. High levels of electrically active substitutional doping using phosphorous also contribute to tensile stress. Preferred precursors for electrical dopants are dopant hydrides, including n-type dopant precursors such as phosphine, arsenic vapor, and arsine. Silylphosphines, for example (H₃Si)_3-xPR_x, and silylarsines, for example, (H₃Si)_3-xAsR_x, where x=0, 1 or 2 and R_x═H and/or deuterium (D), are alternative precursors for phosphorous and arsenic dopants. Phosphor and arsenic are particularly useful for doping source and drain areas of NMOS devices. SbH₃ and trimethylindium are alternative sources of antimony and indium, respectively. Such dopant precursors are useful for the preparation of preferred films as described below, preferably boron-, phosphorous-, antimony-, indium-, and arsenic-doped silicon, Si:C, silicon germanium and SiGe:C films and alloys.

Selective Epitaxial Formation of Tensile Strained Si:C Films.

Techniques have now been developed for selectively forming a tensile strained Si:C film in exposed semiconductor windows, such as recessed source/drain regions of a mixed substrate. For example, it is possible to accomplish such selective formation by (a) blanket depositing a Si:C film over a mixed substrate using trisilane as a silicon precursor, and (b) selectively etching the resulting non-epitaxial layer that is formed over the insulator portion of the mixed substrate. Steps (a) and (b) are optionally repeated cyclically until a target epitaxial film thickness over the recessed source/drain regions is achieved.

It is possible to form recessed source/drain regions by dry etching with subsequent HF cleaning and in situ anneal. In embodiments wherein a dry etch is used, deposition of a selectively grown, thin (between approximately 1 nm and approximately 3 nm) silicon seed layer helps reduce etch damage. A seed layer also helps to cover damage caused by prior dopant implantation processes. In an example embodiment, such a seed layer might be selectively deposited using simultaneous provision of HCl and dichlorosilane at a deposition temperature between about 700° C. and about 800° C.

In accordance with preferred embodiments a cyclical blanket deposition and etch process is illustrated in the flowchart provided in FIG. 1, and in the schematic illustrations of the partially formed semiconductor structures illustrated in FIG. 2 though FIG. 5D. While illustrated for use with Si:C deposition in recessed source/drain regions, it will be appreciated that the techniques described herein are advantageous for selective formation of epitaxial films in other circumstances, such as on active area islands surrounded by field isolation prior to any gate definition and without recessing.

In particular, FIG. 1 illustrates that a mixed substrate having insulator regions and recessed source/drain regions is placed in a process chamber in operational block 10. FIG. 2 provides a schematic illustration of an exemplary mixed substrate that includes a patterned insulator 110 formed in a semiconductor substrate 100, such as a silicon wafer. The illustrated insulator 110, in the form of oxide-filled shallow trench isolation (STI), defines field isolation regions 112 and is adjacent recessed source/drain regions 114 shown on either side of a gate electrode 115 structure. Note that the gate electrode 115 overlies a channel region 117 of the substrate. Together, the channel 117, source and drain regions 114 define a transistor active area, which is typically surrounded by field isolation 112 to prevent cross-talk with adjacent devices. In other arrangements, multiple transistors can be surrounded by field isolation. In one case, the top of the gate structure 115, can be capped with dielectric, as illustrated. This surface then behaves similarly to the field regions 110 with respect to the deposition there over, and the conditions that maintain selectivity in the field region will also apply to the top of the gate. In the case that that the gate 115 is not capped with a dielectric, then the surface of the gate has the potential to grow polycrystalline material which then can be removed through in-situ etching of polycrystalline material, but a different set of selectivity conditions (pressure, gas flow, etc) would apply, compared to those used to ensure no residual polycrystalline material on the field 110.

As indicated by operational block 20 in FIG. 1, and as illustrated schematically in FIG. 3, a blanket Si:C layer is then deposited over the mixed substrate using trisilane as a silicon precursor. This results in amorphous or polycrystalline (non-epitaxial) deposition 120 over oxide regions 112, and lower epitaxial deposition 125 and sidewall epitaxial deposition 130 over the recessed source/drain regions 114. Note that “blanket deposition” means that net deposition results over both the amorphous insulator 110 and the single crystal regions 114 in each deposition phase. While lack of etchant (e.g., lack of halides) is preferred in the blanket deposition, in which case the deposition can also be considered “non-selective,” some amount of etchant might be desirable to tune the ratio of deposited thickness over the various regions, as discussed in more detail below. In case such small amounts of etchant are desirable, the deposition process may be partially selective but nevertheless blanket, since each deposition phase will have net deposition over both the insulator 110 and single crystal region 114.

The regions of amorphous or polycrystalline deposition 120 and the sidewall epitaxial deposition 130 are then selectively etched in an operational block 30 (FIG. 1), thus resulting in the structure that is schematically illustrated in FIG. 4. Preferably, little or no epitaxially deposited Si:C is removed from the lower epitaxial layer 125 in the recessed source/drain regions 114 during the selective etch. As discussed in more detail below, the vapor etch chemistry preferably comprises a halide (e.g., fluorine-, bromine- or chlorine-containing vapor compounds), and particularly a chlorine source, such as HCl Cl₂. More preferably the etch chemistry also contains a germanium source (e.g., a germane such as monogermane (GeH₄), GeCl₄, metallorganic Ge precursors, solid source Ge) to improve etch rates. At the same time that the non-epitaxial material 120 is selectively removed, some epitaxial material is left and some is removed. The sidewall epitaxial layer 130 is of a different plane and is also more defective (due to growth rate differential on the two surfaces) than the lower epitaxial layer 125. Accordingly, the sidewall epitaxial layer 130 is more readily removed, along with the non-epitaxial material 120. Thus, each cycle of the process can be tuned to achieve largely bottom-up filling of the recesses 114. In some arrangements, epitaxial material can be left by the process even on the sidewalls if it is of good quality and does not hinder the goals of the selective fill.

This process is repeated until a target thickness of epitaxial Si:C film thickness is achieved over the recessed source/drain regions 114, as indicated by decisional block 40 (FIG. 1), and as schematically illustrated in FIG. 5A (deposition of second cycle of blanket Si:C layer 120) and FIG. 5B (etch of second cycle of Si:C layer to leave layer of epitaxial Si:C with increased thickness of epitaxial layer 125 in recessed source/drain regions 114). FIG. 5C illustrates the result of further cycle(s) to leave epitaxial refilled source/drain regions 114, where the selective epitaxial layers 125 are roughly coplanar with field oxide 110. FIG. 5D illustrates the result of further cycle(s) to leave epitaxial layers 125 selectively as elevated source/drain regions 114.

The selective formation process may further include addition cycles of blanket deposition and selective etch back from over dielectric regions, but without carbon doping to form a capping layer. The capping layer can be with or without electrical dopants. For example, the portion of the elevated source/drain regions 125 of FIG. 5D that is above the original substrate surface (i.e., above the channel 117) can be carbon-free, since it does not contribute to the strain on the channel 117.

In an example embodiment, the deposited Si:C film optionally includes an electrically active dopant, particularly one suitable for NMOS devices, such as phosphorous or arsenic, thereby allowing phosphorous-doped Si:C films or arsenic doped Si:C films to be deposited (Si:C:P or Si:C:As films, respectively). The Si:C film is preferably deposited with an amorphous-to-epitaxial growth rate ratio that is preferably between about 1.0:1 and about 1.6:1, more preferably between about 1.0:1 and about 1.3:1, and most preferably between about 1.0:1 and about 1.1:1, such that the film thickness over insulator and over the recessed source/drain regions is about equal. Manipulating the amorphous (or more generally non-epitaxial) to epitaxial growth rate ratio advantageously enables manipulation of the facet angle at the interface between the amorphous and crystalline Si:C after the subsequent etch process, and also minimizes etch duration for removal, relative to greater thicknesses over the insulators. Preferably the amorphous regions of the Si:C deposition have little or no crystallinity (i.e., are predominantly amorphous), thus facilitating the subsequent etch in such regions. Furthermore, minimizing the excess of non-epitaxial deposition by bringing the ratio of thickness close to 1:1 reduces the length of the etch phase needed to clear non-epitaxial deposition from the field regions (and optionally from the gate).

In a preferred embodiment, the Si:C film is selectively etched from the mixed substrate using an in situ chemical vapor etching technique. The chemical vapor etching technique is optionally performed simultaneously with a brief temperature spike. In one embodiment, the temperature spike is conducted using the process described in U.S. Patent Application Publication 2003/0036268 (filed 29 May 2002). As described therein, using a single wafer epitaxial deposition tool with radiant heating through cold quartz or otherwise transparent walls, a temperature spike can employ full power to the upper lamps for a short duration (for example, for about 12 to about 15 seconds) while decoupling the power ratio for the lower lamps. In this way, the wafer temperature can rapidly ramp up while the susceptor temperature lags significantly. The wafer temperature preferably increases from the loading temperature by between about 100° C. and about 400° C., and more preferably by between about 200° C. and about 350° C. Because of the short duration of the temperature spike and etch phase, the wafer is allowed to cool before the susceptor gets a chance to approach the peak temperature. In this way, it takes far less time for the wafer to cycle in temperature, as compared to simultaneously cycling the temperature of a more massive combination of wafer/susceptor together. An example reactor for use with this temperature spike technique is the EPSILON® series of single wafer epitaxial chemical vapor deposition chambers, which are commercially available from ASM America, Inc. (Phoenix, Ariz.).

However, in another embodiment, to aid in maintaining high concentrations of substitutional carbon and electrically active dopants, while at the same time minimizing temperature ramp/stabilization times, the etch temperature is preferably kept low. Using a low temperature for the etch also reduces the likelihood that electrically active dopant atoms are deactivated during the etch. For example, etching with Cl₂ gas advantageously allows the etch temperature to be reduced, thus helping to maintain the substitutional carbon and electrically active dopants.

Low temperatures for the etch phase enables roughly matching deposition phase temperatures while taking advantage of the high dopant incorporation achieved at low temperatures. Etch rates can be enhanced to allow these lower temperatures without sacrificing throughput by including a germanium source (e.g. GeH₄ GeCl₄, metallorganic Ge precursors, solid source Ge) during the etch phase instead of flash ramping the temperature to improve throughput.

“Isothermal” cyclical blanket deposition and etching, as used herein, means deposition and etching within ±50° C. of one another, preferably within ±10° C., and most preferably setpoint temperature is within ±5° C. for both steps. Advantageously, isothermal processing improves throughput and minimizes time for temperature ramping and stabilization. Similarly, both blanket deposition and etching process are preferably “isobaric,” i.e., within ±50 Torr of one another, preferably within ±20 Torr. Isothermal and/or isobaric conditions facilitate better throughput for avoiding ramp and stabilization times.

As illustrated in FIG. 1, the two-stage process of performing a blanket deposition followed by a selective etch is optionally repeated cyclically until a target epitaxial film thickness over the recessed source/drain regions is achieved. Example process parameters are summarized in Table A below, which lists both preferred operating points as well as preferred operating ranges in parentheses. As is evident from Table A, the process conditions—such as chamber temperature, chamber pressure and carrier gas flow rates—are preferably substantially similar for the deposition and the etch phases, thereby allowing throughput to be increased. Thus, the example below employs isothermal and isobaric conditions for both phases of the cycle. Other parameters are used in modified embodiments.

	TABLE A
	process phase
				post-		flash-
	stabilize			deposit		bake	post-bake
	chamber			purge of	pre-bake	temp	cool and
	temp and	pre-		process	etchant	spike	temp
TABLE A	atmosphere	deposit	Deposit	gases	stabilize	(optional)	stabilize
time		5	15	5	5	6.5	12.5
(sec)		(2.5-7.5)	(5-20)	(2.5-7.5)	(2.5-7.5)	(3.0-10)	(10-15)
temp	550	550	550	550	550	temp	550
(° C.)	(500-650)	(500-650)	(500-650)	(500-650)	(500-650)	spike	(500-650)
pressure	64	64	64	64	64	64	64
(Torr)	(50-200)	(50-200)	(50-200)	(50-200)	(50-200)	(50-200)	(50-200)
H2/He	2.0	2.0	2.0	2.0	2.0	2.0	2.0
(slm)	(0.5-20)	(0.5-20)	(0.5-20)	(0.5-20)	(0.5-20)	(0.5-20)	(0.5-20)
Cl2/HCl					200	200	200
(sccm)					(5-1000)	(5-1000)	(5-1000)
Si3H8		75	75	75
(mg/min)		(50-200)	(50-200)	(50-200)
CH3SiH3		150	150	150
(sccm)		(10-300)	(10-300)	(10-300)
PH3		50	50	50
(sccm)		(10-200)	(10-200)	(10-200)

Using the parameters provided in Table A, it is possible to achieve net growth rates that are preferably between about 4 nm min⁻¹ and about 11 nm min⁻¹, and more preferably between about 8 nm min⁻¹ and about 11 nm min⁻¹, for epitaxial Si:C:P films that are selectively deposited in recessed source/drain regions. It is also possible to achieve thin Si:C:P films with substitutional carbon content up to 3.6% as determined by Vegard's Law, and with resistivities between about 0.4 mΩ cm and about 2.0 mΩ cm. By manipulating the deposition conditions, it is possible to obtain other film properties.

During the etch process disclosed herein, epitaxial Si:C is etched significantly slower than amorphous or polycrystalline Si:C in each etch phase (etch selectivity in the range of 10:1-30:1). Defective epitaxial material is also preferentially removed in the etch phases. In a preferred embodiment, the cyclical deposition and etch process conditions are tuned to reduce or eliminate net growth on the oxide while achieving net growth in each cycle in the epitaxial recessed source/drain regions. This cyclical process is distinguishable from conventional selective deposition processes in which deposition and etching reactions occur simultaneously.

Tables B and C below give two examples of deposition and etch durations and resultant thicknesses using a recipe similar to that of Table A. The recipes are differently tuned to modulate both deposition and etch rates by increasing the partial pressure of the Si₃H₈ and optimizing etchant partial pressures.

TABLE B
Deposition Phase	Etch Phase
Growth rate [nm/min]	28	13	α-etch rate [nm/min]
Deposition time[s]	22	47.4	Minimum etch time[s]
		60	% overetch
		75.8	Effective etch time[s]
Deposited α-thickness	10.27	16.43	Removed α thickness
[nm]			[nm]
Deposited epi-thickness	9.78	0.82	Removed c thickness
[nm] per deposition step			[nm] per etch step
α/epi growth rate ratio	1.05	20	In-situ etch selectivity
Purge (pre epi + post	25	25	Purge (pre epi + post
epi) [s]			epi) [s]
Final time/cycle[s]	122.8
Final thk/cycle [nm]	8.96
Average growth rate [nm/min]	4.38

TABLE C
Deposition Phase	Etch Phase
Growth rate [nm/min]	80	25	Etch rate [nm/min]
Deposition time[s]	8	25.6	Minimum etch time[s]
		30	% overetch
		33.28	Effective etch time[s]
Deposited α-thickness	10.67	13.87	Removed α thickness
[nm]			[nm]
Deposited epi-thickness	10.67	0.693	Removed c thickness
[nm] per deposition step			[nm] per etch step
α/epi growth rate ratio	1	20	In-situ etch selectivity
Purge (pre epi + post	20	20	Purge (pre epi + post
epi) [s]			epi) [s]
Final time/cycle[s]	61.3
Final thk/cycle [nm]	9.977
Average growth rate [nm/min]	9.76

The process parameters provided in Table A indicate a Cl₂/HCl etch chemistry. In a modified embodiment, between about 20 sccm and about 200 sccm of 10% GeH₄ is included in the etch chemistry as an etch catalyst. In certain embodiments, inclusion of a germanium source (e.g., a germane such as GeH₄, GeCl₄, metallorganic Ge precursors, solid source Ge) in the etch chemistry advantageously enhances the etch rate and the etch selectivity. In addition, use of germanium as a catalyst also advantageously allows lower etch temperatures to be used, and allows a temperature spike during etch to be omitted, as noted above in discussion of isothermal processing. Additional information regarding diffusion of germanium in amorphous, polycrystalline and single crystalline silicon and the subsequent etching of Ge rich silicon materials is provided in the literature; see for example, Mitchell et al., “Germanium diffusion in polysilicon emitters of SiGe heterojunction bipolar transistors fabricated by germanium implantation”, J. of Appl. Phys, 92(11), pp. 6924-6926 (1 Dec. 2002), Wu et al., “Stability and mechanism of selective etching of ultrathin Ge films on the Si(100) surface,” Phys. Rev. B, 69 (2004), and Bogumilowicz et al., “Chemical vapour etching of Si, SiGe and Ge with HCl; applications to the formation of thin relaxed SiGe buffers and to the revelation of threading dislocations upon chlorine adsorption,” Semicond. Sci. & Tech., no. 20, pp. 127-134, (2005).

FIG. 6 shows a graph of etch rate of amorphous regions of a carbon-doped silicon film as a function of HCl partial pressure in the etch chemistry, at a constant temperature of 600° C. By decreasing the H₂ carrier flow, the partial pressure of HCl and GeH₄ is increased, thereby significantly increasing the amorphous etch rate in certain embodiments. For example, FIG. 6 indicates that inclusion of 20 sccm of 10% GeH₄ in the etch chemistry (symbols ▾ and ▴) results in substantially higher amorphous etch rates.

FIG. 7 shows a graph of etch rate and amorphous/epitaxial etch rate ratio as a function of GeH₄ flow in the etch chemistry, at a constant temperature of 600° C., a constant H₂ carrier flow of 2 slm, and a constant chamber pressure of 64 Torr. Amorphous etch rates are indicated by the “a-” prefix in the legend, epitaxial etch rates are indicated by the “c-” prefix in the legend, and etch rate ratios are indicated by “ER” in the legend. Increasing the GeH₄ flow causes the amorphous/epitaxial etch rate ratio to increase to a point, beyond which additional GeH₄ reduces etch selectivity. For example, FIG. 7 indicates that an etch chemistry comprising 200 sccm HCl and approximately 30 to 40 sccm of 10% GeH₄ produces an amorphous/epitaxial etch rate ratio that cannot be obtained with lower or higher GeH₄ flow rates.

FIG. 8 shows a graph of etch rate of amorphous regions of a Si:C film as a function of chamber pressure for various GeH₄ flow rates in the etch chemistry, at a constant temperature of 550° C., a constant H₂ carrier flow of 2 slm, and a constant HCl etchant flow of 200 sccm. By increasing the chamber pressure beyond approximately 80 Torr, the dependence of the etch rate on the GeH₄ flow rate is reduced. However, by increasing the chamber pressure from about 64 Torr to about 80 Torr when 50 sccm of 10% GeH₄ is included in the etch chemistry, the amorphous etch rate is increased by a factor of about two.

FIG. 9 shows a graph of etch rate of amorphous regions of a carbon-doped silicon film as a function of reciprocal temperature, at a constant chamber pressure of 64 Torr, a constant H₂ carrier flow of 2 slm, a constant HCl etchant flow of 200 sccm, and a constant GeH₄ etchant flow of 50 sccm of 10% GeH₄. The absolute etch rates are very high for these chemicals even at very low temperatures.

FIG. 10 shows a graph of thickness of amorphous regions of a carbon-doped silicon film as a function of accumulated etch time, at a constant chamber pressure of 64 Torr, a constant chamber temperature of 550° C., a constant H₂ carrier flow of 2 slm, and a constant HCl etchant flow of 200 sccm. The slopes of the lines plotted in FIG. 10 correspond to the etch rate of the amorphous Si:C film. As indicated, the etch rate in the center of the deposited film is greater than the etch rate at the edge of the deposited film. Therefore, in a preferred embodiment the wafer is “overetched” to increase the likelihood that amorphous Si:C is removed from the slower-etching wafer edges. By extrapolating the lines plotted in FIG. 10 to the y-axis, it is possible to estimate the initial amorphous film thickness and growth rate. Likewise, by extrapolating the lines plotted in FIG. 10 to the x-axis, it is possible to estimate the time required to etch the amorphous material entirely. FIG. 10 illustrates that an etch rate of approximately 140 Å min⁻¹ is obtained with the provided process parameters.

FIG. 11 shows a graph of etch rate of amorphous regions of a carbon-doped silicon film deposited on a wafer as a function of radial position on the wafer, at a constant chamber temperature of 550° C., a constant chamber pressure of 64 Torr, a constant H₂ carrier flow of 2 slm, and a constant HCl etchant flow of 200 sccm. FIG. 11 indicates that the etch rate is slightly slower at the wafer edge than the wafer center.

FIG. 12 shows a graph of thickness of amorphous regions of a carbon-doped silicon film deposited on a wafer as a function of radial position on the wafer for various etch cycle durations, at a constant chamber temperature of 550° C., a constant chamber pressure of 80 Torr, a constant H₂ carrier flow of 2 slm, a constant HCl etchant flow of 200 sccm, and a constant GeH₄ etchant flow of 6.5 sccm.

FIG. 13 shows a graph of thickness of amorphous regions of a carbon-doped silicon film deposited on a wafer as a function of radial position of the wafer for various GeH₄ etchant etch chemistries and various etch cycle durations. As illustrated in FIG. 13, longer etch cycles and higher GeH₄ flow rates leads to more nonuniform etching. In a modified embodiment, this effect is compensated for by providing a final etch cycle of extended duration, thereby providing sufficient “overetch” to remove amorphous Si:C remaining at the center of the wafer. Accordingly, it is desirable to deposit and etch relatively small thicknesses in each cycle, preferably between about 1 nm/cycle and 10 nm/cycle, more preferably between about 2 nm/cycle and 5 nm/cycle. As noted above, conditions similar to Table A have been used to achieve net deposition rates of 4-11 nm/min.

FIG. 14 is a photograph illustrating an example partially formed carbon-doped silicon structure created by performing one deposition cycle and one etch cycle on a patterned substrate. As illustrated, crystalline Si:C:P is present over an epitaxial substrate region, while amorphous Si:C:P is present over oxide. An amorphous pocket is present at the amorphous/epitaxial interface because deposition occurs at different growth rates depending on the exposed crystallographic orientation. In the structure illustrated in FIG. 14, the ratio of amorphous etch rate to epitaxial etch rate is over 20. FIG. 16 is a photograph illustrating an example partially formed carbon-doped silicon structure created by performing multiple deposition and etch cycles on a patterned substrate. As compared to FIG. 14, substantially all amorphous Si:C:P has been removed from oxide surfaces, resulting in pseudo selective epitaxial formation. FIG. 17 illustrates an atomic force microscopy analysis of a epitaxial carbon-doped silicon film that has been selectively deposited using certain of the techniques disclosed herein.

FIG. 15 shows a graph of element concentration as a function of depth for an example partially formed carbon-doped silicon film formed using certain of the techniques disclosed herein. As illustrated, relatively modest levels of germanium are incorporated into the Si:C film due to GeH₄ use during the etch phase. Preferably, the germanium incorporation is less than about 5 atomic %, more preferably less than about 2 atomic %, and most preferably less than about 1 atomic %.

FIG. 18 shows x-ray diffraction rocking curves of carbon-doped silicon films deposited using certain of the techniques disclosed herein. The curves indicate different quantities of deposition/etch cycles, and correspond to increasing monomethylsilane (“MMS”) flow rates, which corresponds to higher substitutional C concentrations in the silicon epitaxial film.

The techniques disclosed herein for selective epitaxial deposition of Si:C films provide several advantages over conventional techniques. For example, cyclical removal of polycrystalline or amorphous Si:C from insulator regions helps to improve the interface between the amorphous Si:C and the epitaxial Si:C. In particular, the cyclical process allows epitaxial growth to occur in interface regions where non-epitaxial growth would otherwise occur. Furthermore, in embodiments wherein the temperature spike during etch is omitted, such that the etch cycle is conducted at a temperature that is equal to, or only slightly elevated from, the deposition cycle, lower temperatures lead to many advantages. Throughput is improved by minimizing temperature (and/or pressure) ramp and stabilization time. Deposition temperatures can still be low enough to achieve high (e.g., 1.0-3.6 at C %) substitutional carbon content, and a large portion of the substitutional carbon and electrically active dopants remain in place during the etch, thus resulting in high substitutional carbon and dopant concentrations in the resulting film.

Several features contribute to the high throughput. For example, use of trisilane has been found to improve deposition rates at very low temperatures, thus minimizing the duration of deposition phases without sacrificing, e.g., high substitutional dopant concentrations as deposited that result from lower temperatures and higher deposition rates. The sequences and choice of precursors also facilitates largely or even wholly amorphous deposition over amorphous insulating regions while also leading to relatively uniform thicknesses (thickness ratios of less than 1.6:1) over both single-crystal and amorphous regions, minimizing overall etch time during the etch phases.

SCOPE OF THE INVENTION

While the foregoing detailed description discloses several embodiments of the present invention, it should be understood that this disclosure is illustrative only and is not limiting of the present invention. It should be appreciated that the specific configurations and operations disclosed can differ from those described above, and that the methods described herein can be used in contexts other than fabrication of semiconductor devices.

Claims

1. A method of selectively forming epitaxial semiconductor material, comprising: providing a substrate with insulating regions and semiconductor windows formed therein;blanket depositing non-epitaxial silicon-containing semiconductor material over the insulating regions and epitaxial silicon-containing semiconductor material over the semiconductor windows by flowing a carbon source vapor with trisilane over the substrate wherein the silicon-containing semiconductor material comprises carbon-doped silicon, and wherein the silicon-containing semiconductor material over the semiconductor windows comprises at least about 1.0% substitutional carbon as-deposited;selectively etching the non-epitaxial, non-single-crystal, silicon-containing semiconductor material from over the insulating regions while leaving at least some epitaxial silicon-containing semiconductor material in the semiconductor windows by exposing the substrate to a halide gas source and a germanium source, wherein the germanium source comprises a germanium etch catalyst;repeating blanket depositing and selectively removing in a plurality of cycles; andwherein blanket depositing and selectively removing produces a net growth rate of the silicon-containing material of between about 4 nm/min and 11 nm/min,wherein in each depositing step, a thickness ratio of non-epitaxial semiconductor material over the insulating regions to the epitaxial silicon-containing semiconductor material over the semiconductor windows is less than about 1.6:1.

2. The method of claim 1, wherein the semiconductor windows comprises recesses below upper surfaces of the insulating regions and repeating comprises filling the recesses with epitaxial silicon-containing semiconductor material.

3. The method of claim 2, wherein depositing comprises filling the recesses with carbon-doped silicon to form recessed source/drain structures.

4. The method of claim 3, wherein depositing further comprises in situ supplying electrical dopants to the recessed source/drain structures.

5. The method of claim 3, wherein depositing further comprises forming a carbon-free capping layer over the carbon-doped silicon.

6. The method of claim 1, wherein the semiconductor windows comprises recesses below upper surfaces of the insulating regions and the epitaxial silicon-containing semiconductor material in the recesses exerts lateral tensile strain on an adjacent region.

7. The method of claim 6, wherein selectively removing further comprises removing silicon-containing epitaxial material from sidewalls of the recesses while leaving silicon-containing epitaxial material at bottoms of the recesses.

8. The method of claim 1, wherein the ratio is between about 1.0:1 and 1.3:1.